[Dylanear]

Reading on SGI Depot it's said that only early Fuels are affected by the failing environmental monitoring ICs. Is there any way to look at the MB revision or system serial numbers and know if a given system will likely suffer from this?

If a system isn't suffering this issue in 2019 is it a good bet it never will? Or can this happen many, many years from manufacture?

[Raion]

They all are at this point.

[jan-jaap]

They all are at this point.

I think it was a little more subtle than that.

The Fuel (and every other system of it's generation, e.g. O3K/O3x0 and Tezro) have environmental monitoring chips which measure temperatures and voltages. They are communicated to the L1 via an I2C bus. Other things like part serial numbers are communicated this was as well.

Early Fuels and O3K systems have a high(er) failure rate in these Dallas I2C chips. That's what the OP is referring to I guess. That's why so many people simply turn off environmental monitoring.

The other Dallas chip is of course the battery backed NVRAM. These have an expected battery life of ~ 10 years so by now they are dead / about to die. But these are easy to replace.

I've got a Fuel CPU here somewhere that's perfectly fine, but because the I2C chip can't communicate it's serial number to the L1, the system won't start.

[Dylanear]

I assume the I2C chips, while made by Dallas aren't a simply pull and plug like the Dallas RTC modules are? My experience with replacing Dallas RTCs comes from Indigo 2s, so perhaps it's not the same module on Fuels.

I have about a dozen SGIs, Indigos, Indys, Indigo 2s, O2s, Octanes, that have been in storage for about 11 years now. At some point I'll have a lot of Dallas modules to replace I suspect. :( I pray that's going to be the worst of it, but who knows what'll happen when I try to power those up again.

[Raion]

Other than the snaphat everything is smd on the Fuel. And it's really a shame. I really wanted to like the fuel but I just can't justify them with how unreliable they have been for me

[Dylanear]

My only complaint about the Fuel before learning of the PSU and monitoring chip issues was it was the case looked like it was built like a cheap PC compared to an Octane. Seems perhaps they tried to save money in multiple ways. :(

Still a fast machine! And while the Tezro may look more like traditional SGI build quality, seems it suffers some of the same issues as the Fuel.

I do love that red color! Makes me miss the Crimson I gave the Computer History Museum. That big ol' red box gave me a thrill every time I walked into my office. Nice clean skins too. Just missing the "Reality Engine" sticker. Then again, talk about unreliable, the hours of tinkering, for the number of times I got the Crimson to boot there were many hours of messing with it. I think I may have only had a X display up once with the VGXT boardset? Never got that Reality Engine working. Booted from a WYSE terminal semi-reliably. Lol.

[Irinikus]

The Tezro doesn't suffer from the same issues as the fuel, it's far more robust, as it's based on a supercomputer node board! (Far more effort was put into it's design!)

The Tezro also has environmental monitoring, but I haven't heard of any cases where this system has failed in a Tezro.

[weblacky]

Pardon my ignorance, but I thought you could replace the DS1780 monitoring chips with new ones in a Fuel? They are still produced (variant), I didn't think they were custom programmed or anything? It's more a soldering challenge, right?

I'm hopefully going to be in the market for fuel and from what I've read in the datasheet the DS1780 monitoring chip cannot tolerate VCC voltages above 5.75v DC. I figured that the failures are due to power supply 5V lines going towards (into) 6v that destroys the chips. Perhaps I'm wrong, there isn't much info.

[Raion]

Pardon my ignorance, but I thought you could replace the DS1780 monitoring chips with new ones in a Fuel? They are still produced (variant), I didn't think they were custom programmed or anything? It's more a soldering challenge, right?

I'm hopefully going to be in the market for fuel and from what I've read in the datasheet the DS1780 monitoring chip cannot tolerate VCC voltages above 5.75v DC. I figured that the failures are due to power supply 5V lines going towards (into) 6v that destroys the chips. Perhaps I'm wrong, there isn't much info.

Nobody has documented efforts to fix due to the fact it is surface mounted and the Fuel was until about 2 years ago cheap enough that you'd just replace the motherboards.

As for it being a power supply problem I don't think that's the case, the capacitors in the Fuel are extremely cheap and the entire layout is just thrown together without much engineering thought going into it. I wouldn't be surprised if a fatal design flaw was overlooked.

I've been persuading people to not try fixing fuels and to part them out instead because they are losing VPros due to insufficient cooling at an exponential rate. This has driven up the market prices, making it harder for other users to keep spares in stock for their Tezro, InfinitePerformance Bricks, and Onyx 350s. While I hate to sacrifice the fuel for new enthusiasts, I also don't want to deal with increasing parts shortage and I don't think anybody else does.

The systems have flaws that are too numerous to overlook and that's why we don't recommend them. They are the DEC Multia of our hobby.

[Irinikus]

There was a member on Nekochan who successfully managed to replace them, if I remember correctly, he was based in Britain. (For the life of me, I can't remember who it was though.)

He had the SMT work professionally done.